letters to nature
Acknowledgements We thank M. Inouye for the gift of the RU1012 strain, L. Loew for the gift of
styryl dyes, S. Conrad and G. Shirman for assistance with mutagenesis and protein chemistry, and
M. G. Prisant for construction of the computer cluster. This work was supported by grants from
the Office of Naval Research, the Defense Advanced Research Project Agency and the National
Institutes of Health.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for material should be addressed to H.W.H.
([email protected]).
..............................................................
Direct observation of catch bonds
involving cell-adhesion molecules
Bryan T. Marshall*, Mian Long*†, James W. Piper*†, Tadayuki Yago‡,
Rodger P. McEverठ& Cheng Zhu*k
* Woodruff School of Mechanical Engineering and k Coulter School of Biomedical
Engineering, Georgia Institute of Technology, Atlanta, Georgia 30332, USA
‡ Cardiovascular Biology Research Program, Oklahoma Medical Research
Foundation, and § Department of Biochemistry and Molecular Biology and
Oklahoma Center for Medical Glycobiology, University of Oklahoma Health
Sciences Center, Oklahoma City, Oklahoma 73104, USA
.............................................................................................................................................................................
Bonds between adhesion molecules are often mechanically
stressed. A striking example is the tensile force applied to
selectin–ligand bonds, which mediate the tethering and rolling
of flowing leukocytes on vascular surfaces1–3 . It has been
suggested that force could either shorten bond lifetimes, because
work done by the force could lower the energy barrier between
the bound and free states4 (‘slip’), or prolong bond lifetimes by
deforming the molecules such that they lock more tightly5,6
(‘catch’). Whereas slip bonds have been widely observed7–14,
catch bonds have not been demonstrated experimentally. Here,
using atomic force microscopy and flow-chamber experiments,
we show that increasing force first prolonged and then shortened
the lifetimes of P-selectin complexes with P-selectin glycoprotein
ligand-1, revealing both catch and slip bond behaviour. Transitions between catch and slip bonds might explain why leukocyte rolling on selectins first increases and then decreases as wall
shear stress increases9,15,16. This dual response to force provides a
mechanism for regulating cell adhesion under conditions of
variable mechanical stress.
Using atomic force microscopy (AFM) (Fig. 1a), we measured the
force dependence of bond lifetimes of P-selectin with two forms of
P-selectin glycoprotein ligand-1 (PSGL-1) or with G1, a blocking
monoclonal antibody (mAb) against P-selectin17 (see Methods).
P-selectin is an extended C-type lectin expressed on activated
endothelial cells and platelets. PSGL-1 is a mucin expressed on
leukocytes. Ca2þ-dependent interactions of P-selectin with PSGL-1
mediate the tethering and rolling of flowing leukocytes on vascular
surfaces in response to infection or tissue injury1–3.
We captured dimeric PSGL-1 purified from human neutrophils18
or monomeric recombinant soluble PSGL-1 (sPSGL-1)19 with PL2,
a non-blocking anti-PSGL-1 mAb20 adsorbed on the cantilever tip
(Fig. 1b). Cantilever tips bearing (s)PSGL-1 or G1 were repeatedly
brought into contact with lipid bilayers reconstituted with P-selectin purified from human platelets21 to allow bond formation. The
cantilever was then retracted a prescribed distance to apply a
constant tensile force to the bond or bonds (if any resulted from
the contact), and the duration or lifetime of the adhesion at that
† Present addresses: National Microgravity Laboratory, Institute of Mechanics, Chinese Academy of
Sciences, Beijing 100080, China (M.L.), and Immucor, Inc., Norcross, Georgia 30091, USA (J.W.P.).
190
force was recorded (Fig. 1c). To measure lifetime at forces lower
than the level of their fluctuations, many instantaneous forces were
averaged (Fig. 1d, e). This enabled the reliable resolution of mean
forces as low as a few piconewtons, and allowed the detected
differences in mean forces to achieve high statistical significance
(Fig. 1f). The binding frequency was kept low (12–20%) to ensure
that most (about 90%) adhesions dissociated as a single step (Fig.
1c, lower tracing). Only single-step dissociations were analysed.
Binding was highly specific. Rendering the cantilever tip functional with (s)PSGL-1 increased adhesion frequencies 3–10-fold
(Fig. 2a and b) and also increased bond lifetimes (Fig. 3a and b).
The inclusion of blocking mAbs against P-selectin (G1) or against
PSGL-1 (PL120) or the divalent-cation chelator EDTA in the
chamber solution decreased adhesion to nonspecific levels.
G1-coated cantilever tips had significantly higher adhesion frequencies and longer bond lifetimes than control PL2-coated tips (Figs 2c
and 3c).
Remarkably, both the P-selectin–sPSGL-1 interaction (Fig. 3a)
and the P-selectin–PSGL-1 interaction (Fig. 3b) exhibited a biphasic
relationship between lifetime and force. The bond lifetimes initially
increased with force, indicating the presence of catch bonds. After
reaching a maximum, the lifetimes decreased with force, indicating
slip bonds. This biphasic pattern was detected in individual experiments with a single cantilever tip on a single bilayer, which included
as few as about five lifetime measurements to calculate a mean and a
standard deviation at each of four force levels to cover the biphasic
region. This pattern remained unchanged as more data were
accumulated (about 400 lifetime measurements for each form of
PSGL-1), which allowed us to examine the distributions of lifetimes.
As with published flow-chamber data7,9,11–13, lifetimes at a given
force seemed to follow an exponential distribution, which was made
linear by plotting ln(number of events with a lifetime of t or more)
Figure 1 AFM system. a, Schematic diagram. b, Making the AFM functional. The
cantilever tip depicted represents a composite of all molecules adsorbed or captured.
sPSGL-1 and PSGL-1 are depicted as monomer and dimer, respectively. c, Force-scan
curves illustrating the cantilever bending (insets) when a compressive or tensile force was
applied to the tip. The upper curves illustrate a contact cycle without binding, where the
retraction curve (from left to right) retraced the approach curve (from right to left). The
lower curves illustrate a contact cycle with binding and lifetime measurement. d, Plot of
instantaneous force against time before (red) and after (blue) an unbinding event.
e, Running means (thick curves) ^ s.e.m. (paired thin curves) against number of data
points. f, P value of the Student’s t-test comparing the difference of the two running
means against numbers of point pairs.
© 2003 Nature Publishing Group
NATURE | VOL 423 | 8 MAY 2003 | www.nature.com/nature
letters to nature
against t; examples are given in Fig. 3d, e. The negative reciprocal of
the slope of the linear fit was used as a lifetime measure and was
plotted against force with the 95% confidence interval as error bar.
For first-order dissociation kinetics, 21/slope equals the mean, ktl,
and standard deviation, j(t), of lifetimes, which were also plotted
against force. The steepness of plots of ln(number of events with a
lifetime of t or more) against t decreased with increasing force in the
ranges that corresponded to the ascending phase of the curves of ktl
against f and of j(t) against f for P-selectin interacting with sPSGL-1
(Fig. 3d) and with PSGL-1 (Fig. 3e). The differences between any
two neighbouring points in these ranges (where most of the data
reside) were statistically significant (P , 0.04), as determined by the
F-test comparing the slopes of the two plots of ln(number of events
with a lifetime of t or more) against t, and confirmed by the lack of
overlaps in the confidence intervals. Thus, the P-selectin–(s)PSGL-1
bonds behaved as catch bonds in the low force range. To our
knowledge, this is the first experimental demonstration of a catch
bond. Our data also indicate that a molecular interaction can
change from a catch bond in one force range to a slip bond in
another force range—a phenomenon that existing models have not
predicted.
We compared the force dependence of lifetimes of the P-selectin–
(s)PSGL-1 bonds with that of the P-selectin–G1 bond in the same
force range. Other assays have revealed significant differences
between selectin–ligand interactions and antigen–antibody interactions that are subjected to tensile force22. In sharp contrast to the
P-selectin–PSGL-1 bond lifetimes (Fig. 3a, b), the P-selectin–G1
bond lifetimes decreased precipitously with force, displaying typical
slip bond characteristics that are well described by the Bell model4
(Fig. 3c). These qualitatively distinct relationships demonstrate
that the catch–slip transitional bond is specific for the P-selectin–
PSGL-1 interaction.
The curves of lifetime against force for the two forms of PSGL-1
had similar biphasic shapes, but the PSGL-1 curve (Fig. 3b) was
shifted relative to the sPSGL-1 curve (Fig. 3a), approximately
doubling the forces and lifetimes. As depicted in Figs 1b and 4a,
native membrane P-selectin and PSGL-1 are dimeric, whereas
sPSGL-1 is monomeric12,19,23. The AFM data suggest that sPSGL-1
forms monomeric bonds with P-selectin, whereas PSGL-1 forms
dimeric bonds with P-selectin, which was further supported by
molecular elasticity measurements (B.T.M., K. K. Sarangapani,
A. Leppänen, R. D. Cummings, R.P.McE. and C.Z., unpublished
observations). A dimeric bond would allow the applied force to be
spread between two subunits, thereby supporting larger forces and
Figure 2 Binding specificity. a, b, The P-selectin bilayer was tested by a cantilever tip with
sequentially changing treatments: contacts were first made with no (s)PSGL-1 on the tip
or no Ca2þ in the medium (open bars); the same tip was then made functional with
sPSGL-1 (a) or PSGL-1 (b), or the medium was changed to include Ca2þ (solid bars);
finally, EDTA or the indicated mAb was added to the medium (hatched bars). c, The
P-selectin bilayer was contacted with a tip adsorbed with either G1 or PL2. Adhesion
frequency for each condition was measured from three to five spots with 50 contacts each
on a single bilayer–tip pair and results are presented as means ^ s.e.m. for at least three
separate bilayer–tip pairs.
NATURE | VOL 423 | 8 MAY 2003 | www.nature.com/nature
shifting the curve rightwards. In addition, the cantilever tip would
not detach from the bilayer until both subunits of the dimeric bond
had dissociated, thereby prolonging the lifetime and shifting the
curve upwards. Indeed, the lifetime data in Fig. 3a, b can be fitted by
the same force dependence of off-rate, if the sPSGL-1 data are
analysed by a model of single-step dissociation of monomeric bonds
and the PSGL-1 data are analysed by a model of two-step dissociation of dimeric bonds (B.T.M., V. Ramachandran, R.P.McE.
and C.Z., unpublished observations). These combined results
strongly suggest that the AFM measured single monomeric
P-selectin–sPSGL-1 bonds and single dimeric P-selectin–PSGL-1
bonds. The ability to differentiate between single monomeric and
dimeric bonds attests to the sensitivity of the lifetime assay and to
the reliability of the catch–slip transitional bond results. Furthermore, it indicates that the measurements reflect lifetimes of the
P-selectin–(s)PSGL-1 bonds, not of the (s)PSGL-1–PL2 bonds or
the P-selectin–bilayer anchors.
With the use of flow chambers, several groups have shown that
lifetimes of tethers of PSGL-1-expressing cells to P-selectin-coated
surfaces decreased with increasing wall shear stress, which is
consistent with slip bond behaviour7,9,11–13. These experiments
were conducted at wall shear stresses ranging from 0.17 to
3.0 dyn cm22, corresponding to tether forces of 20–400 pN. In
addition, AFM was previously used to study P-selectin–PSGL-1
interactions by loading the bonds with increasing forces until
rupture, which suggested a slip bond at forces of more than 50 pN
(ref. 24). Our AFM data revealed that the P-selectin–sPSGL-1 and
P-selectin–PSGL-1 interactions behaved as catch bonds only at
Figure 3 Lifetimes measured by AFM. a–c, Statistical measures of nonspecific mean
lifetimes (black filled squares), mean lifetimes kt l (open blue squares), standard
deviation of lifetimes j(t ) (green triangles) and 21/slope of the plot of ln(number of events
with a lifetime of t or more) against t (red circles), for bonds of P-selectin with sPSGL-1 (a),
PSGL-1 (b) or G1 (c). The nonspecific lifetimes were measured with tips adsorbed with
PL2 only. The curves are trend lines in a and b, but in c it is a Bell model fit, kt l ¼
kt l0 expð2af =k B T Þ; where kt l0 ¼ 4.6 s, a ¼ 0.46 nm, k B is Boltzmann constant, and T
is absolute temperature4. d, e, Plots of ln(number of events with a lifetime of t or more)
against t in the catch bond regime for sPSGL-1 (d) and PSGL-1 (e). The 95% confidence
intervals of the slopes are shown by the paired, colour-matched lines, also shown as error
bars in a–c.
© 2003 Nature Publishing Group
191
letters to nature
forces below 10 and 20 pN, respectively. To address whether catch
bonds could be observed in the flow chamber, we perfused neutrophils or microspheres coupled with sPSGL-1 or G1 at low wall shear
stresses over sparsely coated P-selectin, which is dimeric, or over
recombinant soluble P-selectin (sP-selectin), which is monomeric23
(Fig. 4a). Measurements of transient tether lifetimes confirmed
catch bonds for (s)P-selectin–(s)PSGL-1 interactions in the force
range coinciding with that of the AFM experiments. Both sPSGL-1coupled microspheres and neutrophils displayed a biphasic force–
lifetime relationship (Fig. 4b, left and middle panels). The curves of
21/slope against f were identical for sPSGL-1-coated microspheres
tethering to P-selectin or sP-selectin (Fig. 4b, left panel), as predicted from the inability of sPSGL-1 to form dimeric bonds with
P-selectin. In contrast, the curve of 21/slope against f for neutrophils tethering to P-selectin was shifted rightwards and upwards
relative to that for neutrophils tethering to sP-selectin, approximately doubling the forces and lifetimes (Fig. 4b, middle panel).
These corroborate the AFM data (Fig. 3a, b) and confirm that
neutrophil PSGL-1 formed dimeric bonds with P-selectin but not
with sP-selectin12. The specific nature of the (s)P-selectin–(s)PSGL1 interactions was confirmed by contrasting their behaviour with
that of the G1-coupled microspheres, which exhibited a force
dependence of lifetimes that was consistent only with slip bonds
(Fig. 4b, right panel). The observation of catch–slip transitional
bonds in both AFM and flow-chamber studies strongly suggests that
this counterintuitive behaviour is not an experimental artefact.
P-selectin binds in a stereospecific manner to an amino-terminal
region of PSGL-1 through recognition of tyrosine sulphate residues,
adjacent peptide determinants, and fucose and sialic acid residues
on a properly positioned core-2 O-glycan3,25. The crystal structure
of this complex reveals a broad, shallow binding interface, but it
does not readily explain how force elicits the catch–slip transitional
bond features that we observed26.
The use of both catch and slip bonds might be physiologically
important. Leukocytes must experience a minimum threshold wall
shear stress to tether to and roll on selectins9,15,16. As wall shear stress
is increased, the number of rolling cells first increases and then
decreases in a biphasic pattern similar to our lifetime data. The wall
shear stresses studied were higher than those that evoked catch–slip
transitional bond behaviour. However, multiple bonds might
mediate most leukocyte tethers, which could reduce the force on
individual bonds to the levels that produce the catch–slip bond
transition10,12. Bonds formed at the leading edge of the rolling cell
are gradually loaded by force as they move to the trailing edge. These
bonds will therefore experience forces in the catch bond regime even
at high shear stresses at the wall. At low wall shear stresses, bond
lifetimes would be too brief to support rolling. When a shear
threshold is reached, catch bonds would retard dissociation at the
trailing edge, thereby stabilizing rolling. Catch bonds might explain
why platelet tethering to von Willebrand factor on disrupted vessels
is most effective at higher shear stresses at the arterial wall3. Catch
bonds have also been proposed to regulate flow-dependent bacterial
adhesion27. Transitions between catch and slip bonds might provide
a general mechanism for the precise regulation of cell adhesion
during mechanical stress.
A
Methods
Lifetimes measured by AFM
Figure 4 Lifetimes measured in a flow chamber. a, Schematic diagram showing
microspheres coupled with either sPSGL-1 or G1, neutrophils expressing PSGL-1 on
microvillous tips, and dimeric P-selectin or monomeric sP-selectin adsorbed on plastic.
b, Five sets of data of 21/slope of ln(number of events with a lifetime of t or more) against
t plots were plotted against wall shear stress (bottom abscissa) and force on a tether
(top abscissa) for sPSGL-1 microspheres (left panel), neutrophils (middle panel) and G1
microspheres (right panel) tethering to P-selectin (red squares) or sP-selectin (blue
circles). The colour-matched curves are trend lines through the mean data in the left and
middle panels; the right panel shows a Bell model4 fit, with best-fit parameters
k t l0 ¼ 1.8 s and a ¼ 0.24 nm.
192
Our custom-made AFM (design adapted from V. Moy, University of Miami) used
commercial cantilevers (TM Microscopes, Sunnyvale, CA) with spring constants from 4 to
13 pN nm21, calibrated by the thermal fluctuation method28. PSGL-1 or sPSGL-1 was
captured by anti-PSGL-1 mAb PL2 adsorbed on a tip, which was then blocked in Hanks
buffer with 1% BSA. Anti-P-selectin mAb G1 was adsorbed on a separate tip followed by
blocking. P-selectin was reconstituted in a polyethylenimine (PEI) polymer-supported
lipid bilayer29,30. The PEI cushion reduced nonspecific adhesion by accommodating the
inversely inserted P-selectin. Low molecular densities were used to achieve infrequent
binding. Binding was enabled by using a piezoelectric translator (PZT) to bring the tip to
contact the bilayer for 3 s with a ,20-pN compressive force. Force signals were collected
from the photodiode, which measured the laser reflected on the back of the cantilever. The
cantilever was then retracted at 250 nm s21 and stopped once it had arrived at a
predetermined distance to apply a constant force to the bond or bonds. The lifetime was
measured from the instant when the PZT stopped to the instant of failure of the bond or
bonds.
Force averaging was employed to allow lifetime measurement at forces comparable to
the level of thermal fluctuations (4–10 pN), as illustrated in Fig. 1d–f. The instantaneous
force immediately before unbinding could not be used as the force at which the lifetime
was measured because it was highly variable (see Fig. 1d for an example). However, the
running mean k f l stabilized after a few tens of fluctuating forces were included, and
k f l < 4.86 ^ 0.52 pN was calculated from 100 points for a lifetime of 0.1 s (data collected
at a rate of 1 kHz). Another 100 post-rupture points yielded k f l < 0.00 ^ 0.42 pN. In
both cases, the standard deviations remained at ,5 pN but the s.e.m. decreased with
increasing data points (Fig. 1e). This enabled a reliable resolution of the two mean forces
and allowed the detected differences to achieve high statistical significance, as indicated by
the rapidly decreasing P value of Student’s t-test comparing the difference of the two
running means with increasing numbers of point pairs included in the calculation
(Fig. 1f).
Lifetimes measured at forces ranging from 5 to 75 pN were divided into six to eight
force bins and analysed by three statistical measures: mean lifetime kt l, standard deviation
of lifetime j(t) and 21/slope of the plot of ln(number of events with a lifetime of t or
more) against t. They are related to kinetic parameters by means of models for the
dissociation of a monomeric bond,
k
21
B1 QR
1 þ L1
or of a dimeric bond,
k22
k
21
B2 O R*1 þ L*1 þ B1 QR
2 þ L2
kþ2
where R, L and B denote receptor, ligand and bond, respectively, k þ and k 2 denote onrates and off-rates, respectively, the superscript * indicates the unbound but constrained
leg of a dimeric molecule, and subscripts 1 and 2 denote monomeric and dimeric
molecules or processes, respectively. For lifetimes measured at constant force f, k 21 was
© 2003 Nature Publishing Group
NATURE | VOL 423 | 8 MAY 2003 | www.nature.com/nature
letters to nature
assumed to depend on force but not time. k22 ðf Þ ¼ 2k21 ðf =2Þ because the force on a
dimeric bond is shared equally by the two subunits, each of which behaves as a monomeric
bond and is equally likely to dissociate. k þ2 is assumed to be a constant. The monomeric
model predicts that lifetimes are distributed exponentially. Performing a log–linear plot
renders the exponential distribution linear; thus 2k 21 equals the slope of the ln(number
of events with a lifetime of t or more) against t plot. Also, ktl1 ¼ j1 ðtÞ ¼ 1=k21 : By
comparison, the dimeric model predicts that
ktl2 ¼ ð1 þ kþ2 =k22 Þ=k21 þ 1=k22
24.
25.
26.
and
j2 ðtÞ ¼ {½ð1 þ kþ2 =k22 Þ=k21 þ 1=k22 Š2 2 2=ðk21 k22 Þ}1=2
27.
The k 21(f) values that allow the monomeric model to fit the data in Fig. 3a also allow
the dimeric model to fit the data in Fig. 3b.
28.
Lifetimes measured in a flow chamber
29.
Lifetimes of transient tethers of neutrophils and microspheres were measured in flowchamber experiments as described in refs 12 and 19. Neutrophils were isolated from
healthy donors. Polystyrene microspheres (6 mm diameter; Polysciences, Warrington,
Pennsylvania) either precoated or not with streptavidin were incubated with biotinylated
sPSGL-1 or mAb G1. Neutrophils or microspheres were perfused at various wall shear
stresses over dimeric P-selectin or monomeric sP-selectin adsorbed at low densities on the
chamber floor. The tethers were eliminated by coating the floor with human serum
albumin instead of (s)P-selectin or by the inclusion of the anti-P-selectin mAb G1, the
anti-PSGL-1 mAb PL1 or EDTA in the medium, confirming their specificity. The force f on
the tether of a neutrophil or a microsphere was calculated on the basis of the tether angle,
which was derived after directly measuring the lever arm of the tether by a flow reversal
method19. The conversion factors between wall shear stress and tether force for neutrophils
and microspheres are 112 and 131 pN dyn21 cm22, respectively. Five sets of lifetimes
(,100 tether events in each set) were measured for each interaction at each wall shear
stress. Each set was analysed by the plot of ln(number of events with a lifetime of t or more)
against t, which was fitted by a straight line. The correlation coefficients, R 2, were more
than 0.9 for all 200 fits. The 21/slope values of the fits were plotted against the wall shear
stress (points in Fig. 4b). The difference in mean 21/slope values at any two neighbouring
wall shear stresses was statistically significant (P , 0.05, Student’s t-test), except
occasionally at the beginning of the slip bond regime after transition from the catch bond
regime.
Received 11 November 2002; accepted 19 February 2003; doi:10.1038/nature01605.
1. Vestweber, D. & Blanks, J. E. Mechanisms that regulate the function of the selectins and their ligands.
Physiol. Rev. 79, 181–213 (1999).
2. McEver, R. P. & Cummings, R. D. Perspectives series: cell adhesion in vascular biology. Role of PSGL-1
binding to selectins in leukocyte recruitment. J. Clin. Invest. 100, 485–491 (1997).
3. McEver, R. P. Adhesive interactions of leukocytes, platelets, and the vessel wall during hemostasis and
inflammation. Thromb. Haemost. 86, 746–756 (2001).
4. Bell, G. I. Models for the specific adhesion of cells to cells. Science 200, 618–627 (1978).
5. Dembo, M., Tourney, D. C., Saxman, K. & Hammer, D. The reaction-limited kinetics of membraneto-surface adhesion and detachment. Proc. R. Soc. Lond. B 234, 55–83 (1988).
6. Dembo, M. Lectures on Mathematics in the Life Sciences, Some Mathematical Problems in Biology 51–77
(American Mathematical Society, Providence, Rhode Island, 1994).
7. Alon, R., Hammer, D. A. & Springer, T. A. Lifetime of the P-selectin–carbohydrate bond and its
response to tensile force in hydrodynamic flow. Nature 374, 539–542 (1995).
8. Pierres, A., Benoliel, A. M., Bongrand, P. & van der Merwe, P. A. Determination of the lifetime and
force dependence of interactions of single bonds between surface-attached CD2 and CD48 adhesion
molecules. Proc. Natl Acad. Sci. USA 93, 15114–15118 (1996).
9. Alon, R., Chen, S., Puri, K. D., Finger, E. B. & Springer, T. A. The kinetics of L-selectin tethers and the
mechanics of selectin-mediated rolling. J. Cell Biol. 138, 1169–1180 (1997).
10. Chen, S. & Springer, T. A. An automatic braking system that stabilizes leukocyte rolling by an increase
in selectin bond number with shear. J. Cell Biol. 144, 185–200 (1999).
11. Smith, M. J., Berg, E. L. & Lawrence, M. B. A direct comparison of selectin-mediated transient,
adhesive events using high temporal resolution. Biophys. J. 77, 3371–3383 (1999).
12. Ramachandran, V. et al. Dimerization of a selectin and its ligand stabilizes cell rolling and enhances
tether strength in shear flow. Proc. Natl Acad. Sci. USA 98, 10166–10171 (2001).
13. Chen, S. & Springer, T. A. Selectin receptor–ligand bonds: Formation limited by shear rate and
dissociation governed by the Bell model. Proc. Natl Acad. Sci. USA 98, 950–955 (2001).
14. Merkel, R., Nassoy, P., Leung, A., Ritchie, K. & Evans, E. Energy landscapes of receptor–ligand bonds
explored with dynamic force spectroscopy. Nature 397, 50–53 (1999).
15. Finger, E. B. et al. Adhesion through L-selectin requires a threshold hydrodynamic shear. Nature 379,
266–269 (1996).
16. Lawrence, M. B., Kansas, G. S., Kunkel, E. J. & Ley, K. Threshold levels of fluid shear promote
leukocyte adhesion through selectins (CD62L,P,E). J. Cell Biol. 136, 717–727 (1997).
17. Geng, J.-G. et al. Rapid neutrophil adhesion to activated endothelium mediated by GMP-140. Nature
343, 757–760 (1990).
18. Moore, K. L. et al. The P-selectin glycoprotein ligand from human neutrophils displays sialylated,
fucosylated, O-linked poly-N-acetyllactosamine. J. Biol. Chem. 269, 23318–23327 (1994).
19. Yago, T. et al. Distinct molecular and cellular contributions to stabilizing selectin-mediated rolling
under flow. J. Cell Biol. 158, 787–799 (2002).
20. Moore, K. L. et al. P-selectin glycoprotein ligand-1 mediates rolling of human neutrophils on
P-selectin. J. Cell Biol. 128, 661–671 (1995).
21. Moore, K. L., Varki, A. & McEver, R. P. GMP-140 binds to a glycoprotein receptor on human
neutrophils: evidence for a lectin-like interaction. J. Cell Biol. 112, 491–499 (1991).
22. Chen, S., Alon, R., Fuhlbrigge, R. C. & Springer, T. A. Rolling and transient tethering of leukocytes on
antibodies reveal specializations of selectins. Proc. Natl Acad. Sci. USA 94, 3172–3177 (1997).
23. Ushiyama, S., Laue, T. M., Moore, K. L., Erickson, H. P. & McEver, R. P. Structural and functional
NATURE | VOL 423 | 8 MAY 2003 | www.nature.com/nature
30.
characterization of monomeric soluble P-selectin and comparison with membrane P-selectin. J. Biol.
Chem. 268, 15229–15237 (1993).
Fritz, J., Katopodis, A. G., Kolbinger, F. & Anselmetti, D. Force-mediated kinetics of single P-selectin/
ligand complexes observed by atomic force microscopy. Proc. Natl Acad. Sci. USA 95, 12283–12288
(1998).
Leppänen, A., White, S. P., Helin, J., McEver, R. P. & Cummings, R. D. Binding of glycosulfopeptides to
P-selectin requires stereospecific contributions of individual tyrosine sulfate and sugar residues. J. Biol.
Chem. 275, 39569–39578 (2000).
Somers, W. S., Tang, J., Shaw, G. D. & Camphausen, R. T. Insights into the molecular basis of leukocyte
tethering and rolling revealed by structures of P- and E-selectin bound to SLe(X) and PSGL-1. Cell
103, 467–479 (2000).
Isberg, R. R. & Barnes, P. Dancing with the host; flow-dependent bacterial adhesion. Cell 110, 1–4
(2002).
Hutter, J. L. & Bechhoefer, J. Calibration of atomic-force microscope tips. Rev. Sci. Instrum. 64,
1868–1873 (1993).
McConnell, H. M., Watts, T. H., Weis, R. M. & Brian, A. A. Supported planar membranes in studies of
cell–cell recognition in the immune system. Biochim. Biophys. Acta 864, 95–106 (1986).
Wong, J. Y. et al. Polymer-cushioned bilayers. I. A structural study of various preparation methods
using neutron reflectometry. Biophys. J. 77, 1445–1457 (1999).
Acknowledgements We thank V. Moy for providing the AFM design and training. This work was
supported by grants from the National Institutes of Health.
Competing interests statement The authors declare that they have no competing financial
interests.
Correspondence and requests for materials should be addressed to C.Z.
([email protected]).
..............................................................
Folding at the speed limit
Wei Yuan Yang* & Martin Gruebele*†
* Center for Biophysics and Computational Biology, and † Departments of
Chemistry and Physics, University of Illinois, Urbana Illinois 61801, USA
.............................................................................................................................................................................
Many small proteins seem to fold by a simple process explicable
by conventional chemical kinetics and transition-state theory.
This assumes an instant equilibrium between reactants and a
high-energy activated state1. In reality, equilibration occurs on
timescales dependent on the molecules involved, below which
such analyses break down1. The molecular timescale, normally
too short to be seen in experiments, can be of a significant length
for proteins. To probe it directly, we studied very rapidly folding
mutants of the five-helix bundle protein l 6–85, whose activated
state is significantly populated during folding. A time-dependent
rate coefficient below 2 ms signals the onset of the molecular
timescale, and hence the ultimate speed limit for folding2. A
simple model shows that the molecular timescale represents the
natural pre-factor for transition state models of folding.
The pre-factor n †, together with the activation energy DG †,
determines the rate coefficient k ¼ n† expð2DG† =kTÞ of transition
state theory. For small molecules, n † < kT/h < (0.2 ps)21 is a
reasonable value. Proteins are simply too large to move about and
fold in a fraction of a picosecond. Protein folding pre-factors have
been estimated on the basis of random intramolecular collision
within small peptide loops, yielding n † < 10–100 ns (refs 3–6). This
number might be too small for the molecular timescale of larger
proteins. Their backbone and side chains can collide in more ways
and with more non-additive interactions, introducing additional
roughness into the free-energy landscape. A calculation for the
five-helix bundle protein l 6–85 has yielded numbers closer to 0.5 ms
(refs 7, 8), and evidence from unfolded cytochrome c in denaturant
indicates values in the 1–40-ms range9. Single-molecule experiments
have put an upper limit of 200 ms on the pre-factor10.
This range of pre-factors leaves us with an ambiguity that has so
far precluded the determination of protein folding barriers from
experiment: many combinations of n † and DG † yield the same
© 2003 Nature Publishing Group
193